STEM is an acronym that means a lot to those in the know and probably nothing to the vast majority of the population. However, STEM or Science, Technology, Engineering and Mathematics are where it is at, at least in so far as any nation wanting to improve human wealth and welfare has to have a rich talent in those subjects.
Oxford Mathematics and Computer Science Undergraduate Andreea Marzoca has become joint winner of the The WCIT University IT Awards 2017. The awards recognise outstanding undergraduate and postgraduate IT students within the UK, and were created in 2015 by The Worshipful Company of Information Technologists Charity (WCIT Charity). Criteria for the award included academic excellence, overcoming adversity, entrepreneurial skills and contribution to charity or community.
Gaussian quadrature the Gaussian way
Abstract
Gauss invented Gaussian quadrature following an approach entirely different from the one we now find in textbooks. I will describe leisurely the contents of Gauss's original memoir on quadrature, an impressive piece of mathematics, based on continued fractions, Padé approximation, generating functions, the hypergeometric series and more.
Randomized methods for accelerating matrix factorization algorithms
Abstract
The talk will describe accelerated algorithms for computing full or partial matrix factorizations such as the eigenvalue decomposition, the QR factorization, etc. The key technical novelty is the use of randomized projections to reduce the effective dimensionality of intermediate steps in the computation. The resulting algorithms execute faster on modern hardware than traditional algorithms, and are particularly well suited for processing very large data sets.
The algorithms described are supported by a rigorous mathematical analysis that exploits recent work in random matrix theory. The talk will briefly review some representative theoretical results.
Sampling in shift-invariant spaces
Abstract
Abstract: We study nonuniform sampling in shift-invariant spaces whose generator is a totally positive function. For a subclass of such generators the sampling theorems can be formulated in analogy to the theorems of Beurling and Landau for bandlimited functions. These results are optimal and validate the heuristic reasonings in the engineering literature. In contrast to the cardinal series, the reconstruction procedures for sampling in a shift-invariant space with a totally positive generator are local and thus accessible to numerical linear algebra.
A subtle connection between sampling in shift-invariant spaces and the theory of Gabor frames leads to new and optimal results for Gabor frames. We show that the set of phase-space shifts of $g$ (totally positive with a Gaussian part) with respect to a rectangular lattice forms a frame, if and only if the density of the lattice is strictly larger than 1. This solves an open problem going backto Daubechies in 1990 for the class of totally positive functions of Gaussian type.
Risk-averse optimization of partial differential equations with random inputs
Abstract
Almost all real-world applications involve a degree of uncertainty. This may be the result of noisy measurements, restrictions on observability, or simply unforeseen events. Since many models in both engineering and the natural sciences make use of partial differential equations (PDEs), it is natural to consider PDEs with random inputs. In this context, passing from modelling and simulation to optimization or control results in stochastic PDE-constrained optimization problems. This leads to a number of theoretical, algorithmic, and numerical challenges.
From a mathematical standpoint, the solution of the underlying PDE is a random field, which in turn makes the quantity of interest or the objective function an implicitly defined random variable. In order to minimize this distributed objective, one can use, e.g., stochastic order constraints, a distributionally robust approach, or risk measures. In this talk, we will make use of risk measures.
After motivating the approach via a model for the mitigation of an airborne pollutant, we build up an analytical framework and introduce some useful risk measures. This allows us to prove the existence of solutions and derive optimality conditions. We then present several approximation schemes for handling non-smooth risk measures in order to leverage existing numerical methods from PDE-constrained optimization. Finally, we discuss solutions techniques and illustrate our results with numerical examples.
11:00
Unlikely Intersections in families of elliptic curves
Abstract
What makes an intersection likely or unlikely? A simple dimension count shows that two varieties of dimension r and s are non "likely" to intersect if r < codim s, unless there are some special geometrical relations among them. A series of conjectures due to Bombieri-Masser-Zannier, Zilber and Pink rely on this philosophy. I will speak about a joint work with F. Barroero (Basel) in this framework in the special case of a curve in a family of elliptic curves. The proof is based on Pila-Zannier method, combining diophantine ingredients with a refinement of a theorem of Pila and Wilkie about counting rational points in sets definable in o-minimal structures.
Everyone welcome!
A resting frog can deform the lily pad on which it sits. The weight of the frog applies a localised load to the lily pad (which is supported by the buoyancy of the liquid below), thus deforming the pad. Whether or not the frog knows it, the physical scenario of a floating elastic sheet subject to an applied load is present in a diverse range of situations spanning a spectrum of length scales. At global scales the gravitational loading of the lithosphere by mountain ranges and volcanic sea mounts involve much the same physical ingredients.
The classic picture of how spheres deform (e.g. when poked) is that they adopt something called 'mirror buckling' - this is a special deformation (an isometry) that is geometrically very elegant. This deformation is also very cheap (in terms of the elastic energy) and so it has long been assumed that this is what a physical shell (e.g. a ping pong ball or beach ball) will do when poked. However, experience shows that actually many shells don’t adopt this state - instead, beach balls wrinkle and ping pong balls crumple.
The topology of the Wilsonahedron: A small case study
Abstract
In this talk, I discuss the positive geometry of the Wilson Loop Diagrams appearing in SYM N-4 theory. In particular, I define an algorithm for associating Wilson Loop diagrams to convex cells of the positive Grassmannians. Using combinatorics of these cells, I then consider the geometry of N^2MHV diagrams on 6 points.